An enterprise-grade research engine that synthesizes multi-source data into citation-backed, professional reports using a structured multi-phase pipeline.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install deep-research-bak
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install deep-research-bak using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Deep Research?
The Deep Research skill is a high-fidelity investigative tool designed for AI agents that require more than just simple web lookups. By implementing an autonomy-first approach, this skill allows the agent to conduct complex analysis, identify patterns across diverse sources, and score source credibility. It is an essential addition to the Openclaw Skills ecosystem for users who need comprehensive, professional-grade documentation for technical or business decision-making.
This skill operates through a progressive context management system, ensuring that even large-scale reports maintain coherence and factual integrity. It moves beyond basic information retrieval by incorporating synthesis, critique, and refinement phases, ultimately delivering a polished final product in multiple formats suitable for executive review.
Deep Research Use Cases
- Comparing complex technologies or state-of-the-art reviews.
- Conducting detailed market analysis and forecasting industry trends.
- Multi-perspective investigations for high-stakes enterprise decision-making.
- Generating comprehensive technical reports with 10+ verified citations.
How Deep Research Works
- Scoping and Planning: The agent analyzes the complexity of the request to choose between Quick, Standard, Deep, or UltraDeep modes.
- Retrieval and Triangulation: Information is gathered from multiple sources and cross-verified to ensure data consistency and credibility.
- Synthesis and Outline Refinement: The agent builds a structured outline and synthesizes gathered evidence into high-quality prose.
- Critique and Quality Gates: For deeper research modes, a critique phase identifies gaps or bias which are then addressed in a refinement step.
- Packaging and Delivery: The final report is exported to a dedicated directory in Markdown, HTML (McKinsey style), and PDF formats.
Deep Research Setup
To begin using this skill within your Openclaw Skills environment, ensure you have the necessary scripts and templates configured. The skill relies on Python-based validation and conversion tools.
# Install dependencies and verify citations
python scripts/verify_citations.py --report [path]
# Validate the integrity of the generated report
python scripts/validate_report.py --report [path]
# Convert the final Markdown output to professional HTML
python scripts/md_to_html.py [markdown_path]
Deep Research Data Schema & Taxonomy
The Deep Research skill organizes its output in a structured directory located at ~/Documents/[Topic]_Research_[YYYYMMDD]/. The data hierarchy follows this schema:
| File Type | Purpose | Description |
|---|---|---|
| .md | Primary Source | The core research report containing all findings and the bibliography. |
| .html | Presentation | McKinsey-style formatted report for web viewing. |
| Professional Print | A print-ready version of the synthesized research. | |
| Metadata | Citations | Every claim includes a citation index [N] mapped to a complete bibliography. |
Deep Research Advanced Features
- Multi-mode execution: Choose from four depth levels ranging from 2-minute quick scans to 45-minute ultra-deep investigations.
- Source Credibility Scoring: Automatically filters and prioritizes high-authority sources for technical and enterprise queries.
- Progressive Generation: Supports long-form reports exceeding 18,000 words without losing context or quality.
- Automated Quality Gates: Uses dedicated scripts to ensure no fabricated citations or placeholders exist in the final output.
SKILL.md
Loading
Related Openclaw Skills
HD Video Editing With AI
An AI-powered video editing skill that automates trimming, color correction, and 1080p exporting using high-performance cloud GPUs.
xia-zhuan-audio
A comprehensive audio and video processing toolkit powered by FFmpeg and Whisper AI for conversion, extraction, and automated transcription.
Greenhelix Bundle Trading Bot Suite
A comprehensive 8-guide bundle designed for developers to build, deploy, and manage professional-grade autonomous trading bot fleets.
macOS Desktop Control Enhanced
A powerful suite for programmatic macOS automation, enabling AI agents to control apps, manage processes, and simulate user inputs.
Video Editing for Marketing
An AI-powered video editing tool that transforms raw footage into polished, ready-to-publish marketing videos using cloud-based GPU rendering.
Korean Video Editing With AI
An AI-driven video editing skill that automates Korean subtitling, silence removal, and high-definition cloud rendering for raw footage.
Featured*
